1. Introduction
Government regulations require that lands disturbed by oil sands mining be reclaimed to an equivalent land capability to what existed prior to mining. The equivalent land capability is based on productivity (e.g., supporting diversified vegetation and wildlife), a Land Capability Classification System (LCCS) rating that quantifies nutrient and water availability regimes and other landscape characteristics necessary for government reclamation certification [Cumulative Environmental Management Association (CEMA); CEMA 2006]. Historically, reclamation cover designs were based on assessments of available water-holding capacity (Elshorbagy and Barbour 2007; Huang et al. 2015). More recent industry designs use physically based water dynamics models calibrated using long-term (up to 10 years) monitoring to predict long-term performance using 60-yr historical climate records (Boese 2003; Huang et al. 2011a,b,c, 2015; Keshta et al. 2009; Price et al. 2010; Qualizza et al. 2004).
The long-term performance of reclaimed land disturbed by oil sands mining will be affected by climate change, driven in part by greenhouse gas emissions. Mining and processing of oil sands are associated with 9.3% of these emissions in Canada and about 0.13% globally [Canadian Association of Petroleum Producers (CAPP); CAPP 2016]. Oil sands industry growth since the 1960s has disturbed about 0.02% of Canada’s boreal forests, which serve as the largest terrestrial reservoir of emitted carbon and store almost 22% of the global carbon stocks available on land [Intergovernmental Panel on Climate Change (IPCC); IPCC 2000].
Climate change is expected to intensify the global hydrological cycle because of increases in precipitation and potential evapotranspiration (Huntington 2006). The IPCC Fifth Assessment Report (AR5; IPCC 2013) indicated global average temperatures from 2003 to 2012 increased by 0.78°C compared to 1850–1900; this trend is expected to continue throughout the twenty-first century. Global mean surface temperature is projected to be 1°–3.7°C higher in 2081–2100 compared to 1986–2005, with global mean precipitation likely to increase by 0.5%–4% °C−1 under all scenarios [representative concentration pathway (RCP): RCP2.6, RCP4.5, and RCP8.5]. Scenario RCP8.5 considers rises in CO2 concentrations by the year 2250 to about 2000 ppmv, which is approximately 7 times the preindustrial level. These dramatic changes in climate and atmospheric composition are expected to cause significant changes in global evapotranspiration by the end of the twenty-first century (Pan et al. 2015). Consequently, their impact on water balances must be understood to assess the future performance of reclaimed land.
Water balances for natural sites and reclaimed waste at oil sands mines in northern Alberta, Canada, have been extensively studied. Huang et al. (2011c) and Zettl et al. (2011) conducted numerical modeling and experimental studies to understand mechanisms controlling infiltration and drainage processes in natural, texturally variable, long-term soil–vegetation (SV) monitoring sites near currently operating mines. Huang et al. (2011b) assessed the impact of soil layering (heterogeneous soil texture), climatic variability (historical climatic record), and various vegetation types on plant-available water at a number of SV sites. Probability distributions, similar to those developed by Elshorbagy and Barbour (2007), highlighted the impact of various factors on the magnitude and variability in actual evapotranspiration. Huang et al. (2015) used a calibrated physically based water dynamics model to examine the impact of cover thickness and climate variability on plant-available water at reclamation covers placed over mine waste (sodic–saline shale overburden) over a 60-yr climate cycle. They found median evapotranspiration did not increase significantly for cover thicknesses greater than 80 cm (all other parameters held constant); however, the frequency of freshwater release from these thicker covers (runoff or interflow) dramatically decreased. This work highlighted the need to optimize reclamation cover designs for long-term climate cycles, taking into consideration both plant-available water and the release of freshwater to adjacent surface water bodies. Several studies used system dynamics models to simulate hydrological (Elshorbagy et al. 2007; Keshta et al. 2009) and infiltration and drainage (Huang et al. 2011a) processes for reclaimed soils at Syncrude’s Mildred Lake mine in northern Alberta. However, the potential impact of climate change on oil sands reclamation covers has not been evaluated by linking future climate with physically based simulations of the soil water dynamics at these sites.
Reclamation covers for oil sands mining waste have been conventionally designed so prescribed cover soils and cover depths can provide sufficient plant-available water to support target vegetation over a long-term climate cycle. The climate cycle used in these designs is based on historical monitored climatic conditions (Huang et al. 2015). The purpose here is not to test or evaluate water balance models as applied to soil cover designs for oil sands mining. Rather, we used water balance models that have already been developed, calibrated, and validated at monitored and characterized study sites. These models were adopted without substantive modification, keeping them as simple as possible to highlight the shifts in water balance that may occur if the historical climatic forcing boundary is replaced by a future climate scenario. Various physically based models [e.g., Soil and Water Assessment Tool (SWAT), HydroGeoSphere (HGS)] have been used to assess the impact of climate change on watershed water balance/streamflow (Githui et al. 2009; Leta et al. 2016; Mango et al. 2011) and water movement and availability in the Boreal Plains of Alberta, Canada (Thompson et al. 2017). Keshta et al. (2012) also used a generic system dynamics water balance model to evaluate the performance of alternate cover designs based on future projections of temperature and precipitation from the Third-Generation Canadian Coupled Global Climate Model (CGCM3) under two Special Report on Emission Scenarios (SRES) emission scenarios (IPCC 2000) but did not incorporate an estimation of uncertainty in the simulations. No previous studies have integrated physically based numerical water balance models (particularly, the widely used HYDRUS-1D in oil sands industry) with future climate projections to assess the impacts of climate change on the water balance of oil sands reclamation covers.
This study utilizes climate change projections based on an archive of forcing scenarios released in September 2013 by phase 5 of the Coupled Model Intercomparison Project (CMIP5) of the IPCC (Taylor et al. 2012). The key question being addressed is how future climate change scenarios, updated in September 2013, will alter key indicators of soil cover water balance performance. The indicators chosen are actual evapotranspiration (AET), because it represents the use of available water by vegetation, and net percolation (NP), because, given the assumption of flat-lying covers without runoff, it captures the total release of water (i.e., water yield) from the covers into the underlying waste and ultimately to downgradient surface water receptors. The study also illustrates a methodology by which field-calibrated physically based water balance models can be coupled with future climate change scenarios; this approach does not appear in the literature to date.
The soil profiles used include reclamation covers over a saline–sodic clay shale overburden [30 dump site (D3)] and a reclaimed sand tailings dyke [Southwest Sand Tailings Storage (SWSS)], as well as three natural profiles of glacial fluvial or dune sand (SV10, SV27, and SV60). Both reclamation sites have extensive long-term-monitoring datasets and physically based soil water dynamics models that were adapted for this study. The natural sites (also extensively characterized and modeled) are included to provide a direct comparison to SWSS. Overall, this study expands our understanding of the potential impact of climate change on the long-term performance of oil sands reclamation covers, an issue with no documented consideration by regulators or industry and relatively limited investigation by scientists (Keshta et al. 2012; Rooney et al. 2015; Schneider 2013).
2. Materials and methods
a. Study sites
All soil profiles are located in the boreal mixed-wood ecoregion (Strong and Leggat 1981) (Fig. 1), which has an area of approximately 290 × 103 km2 and a typical prairie climate with mean annual precipitation of 443 mm, most of which occurs in the summer. Mean winter (December–February) and summer (June–August) temperatures are −16.7° and 15.4°C, respectively.
(a) Map of Canada with the province of Alberta, (b) map of Alberta with relevant cities and the broad study area identified, and (c) map of study area with site locations as red squares where (c) is the expanded view of the red box in (b).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
(a) Map of Canada with the province of Alberta, (b) map of Alberta with relevant cities and the broad study area identified, and (c) map of study area with site locations as red squares where (c) is the expanded view of the red box in (b).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
(a) Map of Canada with the province of Alberta, (b) map of Alberta with relevant cities and the broad study area identified, and (c) map of study area with site locations as red squares where (c) is the expanded view of the red box in (b).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
1) Natural sites
The coarse-textured SV sites were established by the industry in 2005 (AMEC Earth and Environmental and Paragon Soil and Environmental Consulting Inc. 2005) for long-term soil–vegetation monitoring and feature a range of soil textures and layering. According to the northern Alberta ecological classification system (Beckingham and Archibald 1996), SV10 and SV27 fall in the “a1” ecosite class (xeric/subxeric moisture regime and a poor nutrient regime characterized by rapid drainage and limited volumes of stored plant-available water following precipitation events) and SV60 falls in the “d2” ecosite class (mesic/subhygric moisture regime and a medium nutrient regime with greater volumes of water stored and available for transpiration).
Soil type and vegetation coverage of these natural sites are provided by Huang et al. (2011c); details of field experiments, sample collection, particle size distribution, and bulk density measurements are provided by Zettl et al. (2011). The SV10 and SV27 natural sites are texturally homogeneous with fine or medium sands, respectively. The natural SV60 profile is texturally heterogeneous with a fine sand layer (45–84 cm) that is overlain and underlain by coarse sand layers. The dominant vegetation types are jack pine (SV10 and SV27) or white spruce, white aspen, black spruce, and balsam fir (SV60). The average sand, silt, and clay fractions among the three natural sites range from 97.4% to 99.2%, from 0.3% to 1.2%, and from 0.5% to 0.9%, respectively.
2) Reclamation covers
The two reclamation sites are located at the Syncrude Canada Ltd. (SCL) Mildred Lake oil sands mine (Fig. 1). The fine-textured D3 reclamation cover has a peat–glacial clay mineral mixture (0–20 cm) overlying a fine-grained glacial soil (20–80 cm) underlain by the overburden clay shale. The SWSS reclamation cover has a very similar peat–mineral mixture (0–45 cm) overlying tailings sand. The dominant vegetation types are trembling aspen and white spruce at D3 and SWSS. The silt and clay size particle fractions in the reclamation covers (e.g., D3) range from 60% to 90% and from 25% to 55%, respectively (Boese 2003).
The reclamation sites have been monitored since 1999 (D3) and 2001 (SWSS). A meteorological tower monitors air temperature, wind speed, radiation, and relative humidity in addition to direct monitoring of the soil cover and underlying mine waste for volumetric water content, suction, and soil temperature. See Barbour et al. (2004), Boese (2003), Huang et al. (2015), and O’Kane Consultants, Inc. (OKC; OKC 2001, 2016) for details of the instrumentation and monitoring program.
b. Climate change projections and downscaling methods
The observed daily precipitation, minimum/maximum/mean temperatures, relative humidity, wind speed, and net radiation for the baseline period (1961–90) were obtained from Environment and Climate Change Canada (ECCC) records at the Fort McMurray Airport station (located ~50–100 km south of the study sites). The future climate change projections (i.e., precipitation, temperature, relative humidity, wind speed, and net radiation) for the baseline (1961–90) and future period (2016–2100) were obtained using the Canadian GCM (CanESM2) combined with three RCPs (RCP2.6, RCP4.5, and RCP8.5; Taylor et al. 2012). Climate change projections for CanESM2 were obtained from the Canadian Centre for Climate Modeling and Analysis (http://www.cccma.ec.gc.ca/data). The datasets from CanESM2 for the time periods 1961–90 and 2016–2100 used in this study include precipitation (pr), daily minimum near-surface air temperature (tasmin), daily maximum near-surface air temperature (tasmax), relative humidity (hur), near-surface wind speed (sfcWind), surface downwelling longwave radiation (rlds), surface upwelling longwave radiation (rlus), surface downwelling shortwave radiation (rsds), and surface upwelling shortwave radiation (rsus).
The GCM meteorological outputs, which have a resolution of 200 km, were downscaled to provide climate specific to the local region (Franczyk and Chang 2009; Hashmi et al. 2011). A widely used and statistically based stochastic weather generator, Long Ashton Research Station Weather Generator (LARS-WG; Racsko et al. 1991; Semenov and Barrow 1997), was used to generate the site-scale time series of future daily precipitation and temperature based on the three RCPs of CanESM2, while daily relative humidity, wind speed, and net radiation were downscaled using a “delta change” or “perturbation” method (Prudhomme et al. 2002). Appendix A provides detailed downscaling methods.
c. Growing season precipitation and potential evapotranspiration
The calculated melt volume was added to any precipitation that occurred during this winter melt period and to stored water within the soil profile at the start of the April–October simulation period.
Average daily temperature for each of the 100 LARS-WG realizations along with relative humidity, wind speed, and net radiation from the delta change downscaling method were used in the Penman–Monteith equation (Allen et al. 1998; Brutsaert 1982) to calculate the 100 realizations of potential evapotranspiration (PET). The observed daily dewpoint temperature (°C) was from ECCC records at the Fort McMurray Airport station. The 100 realizations of daily PET values during the growing season were used in the model as an input variable.
d. Water balance model
The physically based soil water dynamics model HYDRUS-1D, version 4.16 (Simunek et al. 2013), was used to simulate the daily water balance for each site as detailed in Huang et al. (2011b,c, 2015). In this approach, Richard’s equation for transient unsaturated water flow is coupled to a climate/vegetation water flow boundary that incorporates precipitation (rainfall or snowmelt), actual transpiration from root uptake, and actual surface evaporation. Daily water balance components are then calculated from the simulation results, including infiltration into the ground surface, actual transpiration (AT) from the soil profile over the depth of rooting, actual evaporation (AE) from the soil surface, water release from the active rooting zone into the underlying groundwater system (also referred to as NP), and changes in soil water storage (DS) within the rooting zone.
The D3 site model was based on the model presented by Huang et al. (2015) and was developed by calibrating the model against six years of monitoring data (2006–11) from adjacent site D2 and then validating the model against six years of monitoring data (2006–11) from the D3 site. The calibration and validation phase root-mean-square error (RMSE) were 11.1 and 14.4 mm, respectively. Huang et al. (2011b,c) developed the models for the three natural sites (SV10, SV27, SV60) based on detailed site characterization and field testing. Hydraulic parameters for the SV sites were determined from interpretation of full-scale infiltration and drainage tests conducted by Zettl et al. (2011). The mean and standard deviation (SD) values of van Genuchten–Mualem (VG) parameters (see below) and the saturated hydraulic conductivity were obtained from Huang et al. (2011b). Final validation of the models was undertaken by comparing the predicted AET values for the past 60 years to measured tree growth and forest productivity.
The HYDRUS-1D model for SWSS was developed particularly for this study. Cover soils at SWSS are the same as for D3. Monitoring the climate and reclamation soil profile at SWSS has been ongoing since 2001. The model for SWSS was calibrated against 2010 growing season data using the inverse approach (Simunek et al. 2012) and then validated using 2012 growing season data. The model simulated the soil water content in the cover profile with RMSE and R2 values of 0.11 mm and 0.89 in calibration and 0.19 mm and 0.69 in validation, respectively. Hydraulic parameters obtained for the two SWSS cover materials and the three D3 cover materials by Huang et al. (2015) are provided for comparison (see Table B1 in appendix B).
One key difference in these models is that the clay-rich cover soils at the reclamation sites (D3 and SWSS) were characterized using dual-porosity hydraulic functions while the natural profiles (SV sites) were characterized by single-porosity functions. As discussed by Huang et al. (2015), this was due to the presence of a secondary structure of macropore development within the clay-rich cover soils. In both cases, VG equations (van Genuchten 1980) were used to describe these functional relationships for water storage and hydraulic conductivity. These equations and their parameter constants are defined in appendix B (Table B2 summarizes the parameter values used for the various SV profile simulations).
The modeling domains at each site were as follows: three layers (i.e., peat–mineral soil mixture, secondary clay cover layer, and underlying overburden shale) at the reclaimed D3; two layers (i.e., peat–mineral mixture and tailings sand) at the reclaimed SWSS; and 14, 20, and 18 layers of varying texture and bulk density at SV10, SV27, and SV60, respectively. The large number of layers at the SV sites was recommended by Huang et al. (2011c, 2015) to capture the impact of the observed textural layering on the water balance.
Current industry practice tracks reclamation cover water balances based on monitoring and modeling in one dimension. All previously published models were for flat-lying site profiles, and these conditions were retained here. To make the comparison between the models as simple as possible, we also assumed negligible runoff. This is consistent with the original model development for natural sites that feature well-drained profiles on relatively flat ground. SWSS is also located on similar well-drained sand tailings on flat ground. D3 has both a plateau site with minimal runoff, as well as sloping covers with an average 34 mm of runoff each year (Huang et al. 2015). Although “real” landscapes are often strongly influenced by the redistribution of water in three dimensions, limiting the models to 1D cases (without runoff) simplified comparison of the performance of a range of cover scenarios under changing climatic conditions. Interpreting all profiles as being on flat-lying sites also eliminates differences resulting from runoff. Excess water (i.e., runoff due to a slope) is largely incorporated in the NP values; all NP eventually reports to downgradient surface water bodies, which can be viewed as parallel to water yield.
All models (as adopted from previous studies) utilized a free drainage (i.e., unit vertical gradient) lower boundary because the water table at all sites is deep and consequently the soil water balance is decoupled from the surface water balance (see Dobchuk et al. 2013). The upper boundary is represented by precipitation (rain or snowmelt) or PET obtained from the Penman–Monteith equation. Rainfall interception was estimated using the Braden (1985) equation. PET is distributed between potential evaporation (PE) and potential transpiration (PT) based on a specified leaf area index (LAI) as per Feddes et al. (1974). PT is distributed across a prescribed root depth based on the prescribed root distribution. AE from the surface is some fraction of PE based on a limiting suction at the ground surface. AT is similarly calculated from PT and some limiting suction over the root depth. The root distributions used in the model are from Huang et al. (2011b, 2015). Simunek et al. (2013) provide a detailed description of this modeling approach.
A set of four different climatic inputs were then used with all models to simulate the long-term hydrological performance of the covers. The climate datasets include historical meteorological monitoring data (1961–90) and three future climate projections.
e. Coupling of AET and LAI
The assumed seasonal variation in LAI over the growing season was represented as shown in Fig. 2. Literature-based relationships describing the minimum required AET to support a particular value of LAI were used to constrain the simulations, similar to Huang et al. (2011b). These relationships link LAI, aboveground net primary production (ANPP), and AET. However, the linear relationship between ANPP, LAI, and AET varies depending on the plant species (see Huang et al. 2011b).
Variation in assumed LAI values during the growing season for all five study sites (adapted from Huang et al. 2015).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Variation in assumed LAI values during the growing season for all five study sites (adapted from Huang et al. 2015).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Variation in assumed LAI values during the growing season for all five study sites (adapted from Huang et al. 2015).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Typical values of LAI at the D3, SWSS, SV10, SV27, and SV60 sites are 4.0, 3.0, 1.5, 1.6, and 2.4, respectively (Barr et al. 2012; Huang et al. 2011b, 2015; OKC 2007). To ensure the LAI was appropriate for future climate, simulations were undertaken using a range of LAI values. The AET values from all simulations were then plotted along with the LAI–AET relationship from the literature (see section 3c) to identify the maximum sustainable LAI at each site using the literature-based LAI–ANPP–AET relationships (i.e., the value at which the simulated AET and literature relationship between LAI and AET intersect).
To compare changes in the water balance components, a single maximum value of LAI, obtained from baseline climatic conditions, was used for each site and climate scenario. However, potential shifts in LAI were evaluated at each site for each future climate scenario. A more detailed analysis is included in section 3c (see Table 2, Fig. 7).
3. Results and discussion
a. Downscaling performance during the baseline period
The downscaling method (LARS-WG) provided a similar set of climatic observations to the measurements at the ECCC station over the validation period (1991–2011; Fig. 3), and the two were compared. The thick black line dividing the boxplots in Fig. 3 represents the median value of the distribution. Each box ranges from the 25th to 75th percentiles of the distribution [e.g., the interquartile range (IQR)]. The whiskers further extend to 1.5 IQR from both ends of the box. The observations used for comparison included mean daily precipitation; mean extreme precipitation; maximum of the extreme precipitation; variance of daily precipitation; proportion of dry days (i.e., with zero precipitation) in each month; and annual, growing season, spring, summer, fall, and winter precipitation.
Performance of LARS-WG based on the observed monthly (solid lines) and seasonal (circles) properties and 100 realizations of synthetic (boxplots) precipitation time series at Fort McMurray Airport station for the validation period (1991–2011). (a) Mean precipitation, (b) mean of extreme precipitation, (c) maximum of the extreme precipitation, (d) variance of daily precipitation, (e) proportion of dry days for each month, and (f) total annual and seasonal precipitation amounts. The boxplots depict the range of IQR values (median shown as thick black line) for each scenario with 100 simulations, with whiskers representing values within 1.5 IQR extending from both ends of the boxes and red markers outside the whiskers representing outliers.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Performance of LARS-WG based on the observed monthly (solid lines) and seasonal (circles) properties and 100 realizations of synthetic (boxplots) precipitation time series at Fort McMurray Airport station for the validation period (1991–2011). (a) Mean precipitation, (b) mean of extreme precipitation, (c) maximum of the extreme precipitation, (d) variance of daily precipitation, (e) proportion of dry days for each month, and (f) total annual and seasonal precipitation amounts. The boxplots depict the range of IQR values (median shown as thick black line) for each scenario with 100 simulations, with whiskers representing values within 1.5 IQR extending from both ends of the boxes and red markers outside the whiskers representing outliers.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Performance of LARS-WG based on the observed monthly (solid lines) and seasonal (circles) properties and 100 realizations of synthetic (boxplots) precipitation time series at Fort McMurray Airport station for the validation period (1991–2011). (a) Mean precipitation, (b) mean of extreme precipitation, (c) maximum of the extreme precipitation, (d) variance of daily precipitation, (e) proportion of dry days for each month, and (f) total annual and seasonal precipitation amounts. The boxplots depict the range of IQR values (median shown as thick black line) for each scenario with 100 simulations, with whiskers representing values within 1.5 IQR extending from both ends of the boxes and red markers outside the whiskers representing outliers.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The downscaled daily precipitation ranges from 0 (5th percentile) to 16.6 mm (99th percentile) compared to observed values of 0 (5th percentile) and 16.2 mm (99th percentile). The downscaling method produced minimum daily temperatures ranging from −27.3° (5th percentile) to 14.2°C (99th percentile), which are comparable to measured values of −30.9° (5th percentile) and 14.6°C (99th percentile). The downscaled maximum daily temperatures range from −16.4° (5th percentile) to 29.6°C (99th percentile) compared to observed values of −18.7° (5th percentile) to 30.7°C (99th percentile). Examination of the downscaled values, calculated by averaging the 100 realizations of precipitation and temperature from LARS-WG, indicates extreme precipitation and temperature values, as well as percentile values of daily mean temperature (Table 1) are simulated reasonably well (relative errors < 10% in most cases).
Percentiles of measured and simulated daily precipitation (mm day−1) and temperature (°C) by LARS-WG with relative error (%) at Fort McMurray Airport station during the validation period (1991–2011).
b. Projected changes in temperature, precipitation, and PET
The probability distributions of temperature for the Fort McMurray Airport station during the baseline period (1961–90) and the three climate change scenarios (CanESM2 with RCP2.6, RCP4.5, and RCP8.5) for 2016–40, 2041–70, and 2071–2100 were compared (Fig. 4). The results indicate summers are projected to be hotter, as extreme maximum temperature (99th percentile) increases in the future (37.2°C for RCP8.5 during 2071–2100) compared to the baseline case (26.2°C); winters are also projected to be warmer, with the corresponding extreme minimum temperature (5th percentile) increasing from baseline (−39.4°C) to 2071–2100 (−24.3°C) for RCP8.5. All RCPs indicate increasing temperature during the twenty-first century, with RCP8.5 (2071–2100) showing the greatest increase of 138% (median value) and RCP2.6 (2071–2100) and RCP8.5 (2016–40) showing the least increase of 28% (median value) compared to the baseline period.
Distribution of daily mean temperature based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline daily mean temperature. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of daily mean temperature based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline daily mean temperature. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of daily mean temperature based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline daily mean temperature. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The projected growing season (April–October) precipitation [i.e., snow water equivalent (SWE) and rainfall] in each year was also plotted (Fig. 5). During the baseline period, the 95% confidence interval (CI) for SWE ranges from 40 to 122 mm and for growing season rainfall ranges from 212 to 483 mm. SWE either shifts upward or downward during the twenty-first century and ranges from 39 to 129 mm depending on the RCP and time period. Rainfall shifts upward during the twenty-first century and ranges from 231 to 581 mm for all RCPs. RCP8.5 (2071–2100) shows maximum annual increases in median SWE and growing season rainfall of 9.4% and 22.8%, respectively. However, RCP4.5 (2041–70 and 2071–2100) and RCP8.5 (2016–40) show annual decreases in median SWE, and RCP2.6 (2016–40) shows the minimum increase of 7.4% in growing season median rainfall compared to the baseline period. Some scenarios (e.g., RCP4.5 for 2071–2100) feature a downward shift in median SWE compared to the baseline period but a significant upward shift in growing season median rainfall. These shifts for the same RCP and time period might be due to a warmer winter and extended growing season in the future compared to the baseline period.
Distribution of growing season (April–October) (top) SWE, (middle) rainfall, and (bottom) sum of the two based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed lines represent baseline median SWE, rainfall, and SWE + rainfall values. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of growing season (April–October) (top) SWE, (middle) rainfall, and (bottom) sum of the two based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed lines represent baseline median SWE, rainfall, and SWE + rainfall values. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of growing season (April–October) (top) SWE, (middle) rainfall, and (bottom) sum of the two based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed lines represent baseline median SWE, rainfall, and SWE + rainfall values. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Generally, the future increase in annual precipitation (with the exception of RCP4.5 during 2041–70) is more intensified toward the end of the twenty-first century. Multi-GCM ensembles might be useful to further investigate the irregularity in RCP4.5. A similar behavior (without any exception for RCP4.5) was observed for extreme precipitation intensities in Saskatoon, Saskatchewan, Canada, based on precipitation projections from eight GCMs, three RCPs, and two downscaling methods (Alam and Elshorbagy 2015). Srivastav et al. (2014) also noted increased rainfall intensities during the twenty-first century at four rainfall stations in Canada, while Hassanzadeh et al. (2014) observed increases in short-duration annual maximum precipitation in Saskatoon for all RCPs/emission scenarios. Thompson et al. (2017) showed predicted annual precipitation, based on 13 climate change scenarios, would increase for all scenarios by the end of the twenty-first century at a catchment within the Boreal Plains of northern Alberta. Suncor Energy, Inc. (2007), found annual precipitation would increase in the Fort McMurray region in 2041–69 compared to 1961–90 using Canadian Global Coupled Model, version 2 (CGCM2), and two emission scenarios (A2 and B2); they also used other GCMs, and almost all showed increases in future annual precipitation in the Fort McMurray region.
The distribution of growing season PET is plotted in Fig. 6. The 95% CI ranges from 525 to 630 mm during the baseline period, increasing to 553–695 mm for all RCPs and time periods. Compared to the baseline period, RCP8.5 (2071–2100) had the largest increase (11.9%) in growing season PET, and RCP4.5 (2016–40) had the smallest (3.3%). Growing season PET increases in all future scenarios compared to the baseline period and, generally, intensifies more toward the end of the twenty-first century with the exception of RCP2.6, which is not unexpected as the temperature would peak before the twenty-first century (e.g., 2050) and then decline for RCP2.6 (Rogelj et al. 2012). An increased projection of annual PET rate was also found in other parts of Canada (Kienzle et al. 2012; Schindler and Donahue 2006).
Distribution of growing season (April–October) PET based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline growing season median PET. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of growing season (April–October) PET based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline growing season median PET. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Distribution of growing season (April–October) PET based on 100 simulations from LARS-WG during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The horizontal gray dashed line represents baseline growing season median PET. Description of the boxplots is as per Fig. 3.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
c. Maximum sustainable LAI and evolution under climate change
The maximum sustainable LAI values (e.g., where simulated AET lines intersect the threshold AET line) at SWSS during the baseline (1961–90) and three future (2016–40, 2041–70, and 2071–2100) periods are 3.8, 4.2, 4.6, and 4.7, respectively. The respective values for SV10 are 1.9, 2.1, 2.2, and 2.3 (see Fig. 7). The threshold AET value was obtained from the relationship between the threshold AET and ANPP (Rosenzweig 1968), where the ANPP was calculated based on its linear relationship with LAI using the LAI and ANPP measurements for all evergreen tree species (Chasmer et al. 2008; Hall et al. 1995; Howard et al. 2004; Lavigne et al. 2005; Vogel and Gower 1998).
Maximum LAI that can be supported by reclamation cover SWSS and natural site SV10 during the baseline (1961–90) and future (2016–2100 based on RCP8.5) periods.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Maximum LAI that can be supported by reclamation cover SWSS and natural site SV10 during the baseline (1961–90) and future (2016–2100 based on RCP8.5) periods.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Maximum LAI that can be supported by reclamation cover SWSS and natural site SV10 during the baseline (1961–90) and future (2016–2100 based on RCP8.5) periods.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The median baseline and future LAI values based on the simulated AET are shown in Table 2; values based on RCP8.5 are consistent with those in Fig. 7. Notably, the partition of PET into PE and PT is no longer controlled by LAI at LAI values above 2.7 (Huang et al. 2011b), which is also apparent in Fig. 7. As a consequence, water balance components for the future climate scenarios presented below are, for consistency, based on the baseline LAI value for each site. However, the estimation of LAI using the simulated AET values results in slightly different LAI values at each site for each time period. The LAI value (Table 2, first row) was used for both the baseline and future time periods; using a single LAI value ensured the simulated AET and NP reflect the impact of changes in climate without further coupling to changes in LAI. Overall, all soil profiles could support marginally higher LAI values in the future, with plant physiological changes taking place over time, compared to the baseline period irrespective of the RCP, time period, and soil profile, with more increase evident toward the end of the twenty-first century.
The simulated median LAI for all scenarios and soil profiles.
d. Baseline water balance components
The long-term (1961–90) growing season (April–October) median water balance components, precipitation (rainfall and SWE), AET (AT and AE), NP (percolation and runoff), and DS, are shown in Table 3. The median precipitation at the Fort McMurray Airport station was 407 mm during the growing season with a median rainfall of 331 mm (more than 80% of total precipitation) and a median SWE of 76 mm (less than 20% of precipitation). The proportion of AET to precipitation at five study sites varies from 62% at SV10 to 94% at SWSS where the proportion of AT to AET varies from 69% to 91%. These two proportions (i.e., AET:precipitation and AT:AET) are higher at the reclamation covers (with higher LAI values) than the natural SV sites. Similarly, the proportion of NP to precipitation varies from 6% at SWSS to 38% at SV10, reflecting the higher NP at the natural SV sites.
Median values of the water balance components during the baseline period (1961–90) at five study sites.
e. Baseline and future water balance
The reclamation covers have higher AET values than natural sites because of higher available water storage. Among all sites, SWSS has the highest AET values (i.e., 383 mm), while SV10 had the lowest (i.e., 253 mm) during the baseline period (Table 4). Of the natural sites, AET values for SV60 are the highest (i.e., 370 mm) and closer to values for the reclamation covers than other two natural sites (SV10 and SV27), while AET values for SV10 and SV27 (253 and 302 mm, respectively) are closer to each other. The reclamation cover SWSS is more similar to SV60 than other SV sites with respect to the AET values (383 and 370 mm, respectively).
Mean/median (SD) of the annual/growing season precipitation (mm), evapotranspiration (mm), NP (mm), and LAI at all study sites compared to other work conducted on the southern boreal forests of western Canada. Asterisks refer to Barr et al. (2012).
This pattern of behavior is consistent with the concept of available water-holding capacity (AWHC; volume of water stored between field capacity and wilting point over the depth of the cover, or rooting depth in the case of natural profiles). AWHC is used in the industry to compare natural and reclaimed profiles as part of reclamation cover design. Likewise, AWHC is not used here as a modeling parameter but rather as an index to compare changes in the performance of various sites as a result of climate change and to allow the covers to be grouped based on similar responses to climate change.
AWHC values (for assumed/known cover depth of 100 cm) for SWSS and SV60 are 267 and 146 mm, but for SV10 and SV27 are 74 and 121 mm, respectively (Huang et al. 2011b; Zettl et al. 2011). These values are consistent with the estimated sustainable LAI values for these sites (Table 2), with LAI values for SV10 and SV27 being similar and lower (1.9–2.5) and for SWSS and SV60 being similar and much higher (3.6–3.8). The reclamation cover (D3) has an AWHC value of 377 mm and an LAI value of 3.8.
The simulated LAI, AET, and NP for the five study sites during the baseline period were compared to those for a range of boreal forest sites determined by Barr et al. (2012; Table 4 herein) to place the simulated water balances for the study sites within the context of sites in the same region. Despite the differences in site vegetation and soil texture, some clear similarities between the boreal forest sites and the simulations are evident. Of the five sites, D3 and SWSS (reclamation covers) and SV60 (d2 ecosite) most resemble the Old Aspen site from Barr et al. (2012) with respect to ET, NP, and LAI values, while SV10 and SV27 (a1 ecosites) resemble the Old Jack Pine and Old Black Spruce sites.
The reclamation covers have the lowest NP values during the baseline period compared to the SV sites (Table 4), consistent with their higher AET values and consequently less water released to NP (Straker et al. 2014). The SV sites produce greater NP because they are unable to store and utilize the available water for growing season AET. Among the natural sites, SV60 has the lowest NP during the baseline period, while SV10 has the highest. NP rates for reclamation covers D3 and SWSS are closer to each other than to the natural sites. Of the natural sites, the NP rates for SV60 are closer to the rates of the reclamation covers than the other two natural sites (SV10 and SV27), while the NP rates of SV10 and SV27 are closer to each other.
Median growing season precipitation and AET and NP values for all sites and simulation time periods are summarized in Table 5, respectively. The growing season median precipitation, AET, and NP increase (shown in parentheses) for all RCPs and future time periods compared to the baseline period; these changes are more pronounced toward the end of the twenty-first century. Considering all RCPs, the maximum increase in precipitation and AET for all sites is approximately 20%–25% and 12%–14%, respectively. All study sites show greater increases in precipitation and AET toward the end of the twenty-first century regardless of RCP. However, because of the relatively low NP rates from sites with the highest water storage (i.e., reclamation covers), their relative changes in NP rates are actually greater (195% for SWSS, 149% for SV60, 138% for D3, 62% for SV27, and 46% for SV10 for 2071–2100) than most of the SV sites, taking all three RCPs into account.
Growing season median precipitation, AET, NP, and DS (mm) for all scenarios and soil profiles. Percentage increase/decrease in future median precipitation, AET, NP, and DS compared to the respective baseline median precipitation, AET, NP, and DS is shown in parentheses.
The simulated AET and NP values for the five study sites in the 2041–70 climate period for the three RCPs are compared to the baseline period in Figs. 8 and 9, respectively. The variability in AET and NP during the future periods is similar to the baseline period, as illustrated by the parallel probability distributions. The magnitude of the growing season AET and NP and corresponding shifts in future AET and NP compared to the baseline period are dependent on the selection of RCP and future time period at each of the study sites. The multiple (100) realizations of LARS-WG, as input to the soil–vegetation–atmosphere transfer (SVAT) model, encompass the uncertainty in the water balance components because of natural climate variability. Overall, the study sites are expected to have increased AET and NP rates (see Table 5 and Figs. 8, 9) regardless of which RCP and time period are considered from the Canadian GCM (CanESM2).
Probability distribution of growing season (April–October) AET in each soil profile based on 100 simulations from HYDRUS-1D during the baseline (1961–90) and future (2041–70) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The future time period 2041–70 is only shown as an example. The horizontal gray dashed line represents the growing season median AET for all cases.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Probability distribution of growing season (April–October) AET in each soil profile based on 100 simulations from HYDRUS-1D during the baseline (1961–90) and future (2041–70) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The future time period 2041–70 is only shown as an example. The horizontal gray dashed line represents the growing season median AET for all cases.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Probability distribution of growing season (April–October) AET in each soil profile based on 100 simulations from HYDRUS-1D during the baseline (1961–90) and future (2041–70) periods using three scenarios (RCP2.6, RCP4.5, and RCP8.5) of CanESM2. The future time period 2041–70 is only shown as an example. The horizontal gray dashed line represents the growing season median AET for all cases.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
As in Fig. 8, but for NP.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
As in Fig. 8, but for NP.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
As in Fig. 8, but for NP.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
It is clear from Figs. 8 and 9 that the probability of exceeding the baseline median AET and NP at each of the five study sites will increase from 50% to approximately 80% for RCP8.5 in 2041–70. This higher frequency of elevated AET is expected to be reflected in more rapid vegetation growth and possibly a shift in dominant species at each site. The increase in NP could have both a positive and negative impact. It would increase the water yield from these landforms, which might support receiving wetlands. However, it could also lead to more rapid flushing and release of chemical constituents of concern from the underlying mine waste.
f. Seasonal water balance
The long-term median seasonal cycle of water balance components (i.e., precipitation, AET, and NP) over the growing season is shown in Fig. 10. The median monthly precipitation at the Fort McMurray Airport station peaks in April (as snowmelt starts at the beginning of the spring season) and also in July (coinciding with the summer rainfall). This pattern is the same for the future period (2041–70) as it was for the baseline. Although all RCPs show higher precipitation in April than the baseline case, RCP8.5 shows lower precipitation in July than the baseline case. This could be due to increased precipitation in the winter and decreased precipitation in the summer for RCP8.5. May, September, and October seem to be the months with lower precipitation compared to other months. In general, the long-term median monthly AET is higher in May, June, and July than other months for both the baseline, as well as the RCPs, consistent with the seasonal variation of LAI (shown in Fig. 2). Although Fig. 8 shows increased AET rates for all RCPs compared to the baseline case, the monthly AET might be lower than the baseline case for some RCPs as shown in Fig. 10. In general, the long-term median monthly NP is the highest in April for both the baseline and the RCPs. According to the probability distribution of NP in Fig. 9, all RCPs show increased NP in the future compared to the baseline NP rates; however, the median monthly NP might be lower in the future than the baseline NP in some months.
Seasonal variation in the water balance components (a) monthly precipitation at the Fort McMurray Airport station, (b) monthly AET at five study sites, and (c) monthly NP at five study sites. All the above plots show water balance components for baseline (1961–90), as well as three RCPs (during 2041–70) at the five study sites. The line styles for the baseline and three RCPs in (b) and (c) follow the corresponding line styles in (a).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Seasonal variation in the water balance components (a) monthly precipitation at the Fort McMurray Airport station, (b) monthly AET at five study sites, and (c) monthly NP at five study sites. All the above plots show water balance components for baseline (1961–90), as well as three RCPs (during 2041–70) at the five study sites. The line styles for the baseline and three RCPs in (b) and (c) follow the corresponding line styles in (a).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Seasonal variation in the water balance components (a) monthly precipitation at the Fort McMurray Airport station, (b) monthly AET at five study sites, and (c) monthly NP at five study sites. All the above plots show water balance components for baseline (1961–90), as well as three RCPs (during 2041–70) at the five study sites. The line styles for the baseline and three RCPs in (b) and (c) follow the corresponding line styles in (a).
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
g. Uncertainty in the simulated water balance components
The simulated 100 realizations of water balance components (AET and NP) were used to demonstrate uncertainty in the simulated median AET and NP due to natural variability. This is represented by the coefficient of variation (CV) values for the corresponding baseline period and future time periods and RCPs for each of the study sites (Table 6). The calculated CV values show that variability in the simulated AET and NP due to natural climate variability may not change significantly in the future compared to the baseline case, where the uncertainty due to natural variability is not attributable to climate change (Alam and Elshorbagy 2015). Moreover, uncertainties due to GCMs are most important in future projections of climate variables (Najafi et al. 2011). Therefore, more focus was given here to uncertainty in future water balance components due to GCMs and associated RCPs and future time periods.
CV of the simulated 100 realizations of growing season median AET and NP for the baseline period, as well as future periods and RCPs at each study site.
h. Changes in the soil water storage
Importantly, net changes in annual water storage will be small relative to other aspects of the water balance over a long climate cycle, for which, “on average,” the change in water storage tends to zero. However, when the climate is more variable and nonstationary, determining if changes in water storage dynamics are reflected in the various climate change scenarios is of value. Mean changes in soil water storage of the five study sites were calculated over a 30-yr climate cycle for the baseline and future time periods based on three RCPs and three future time periods (Table 5). DS would decrease in the future for the D3 reclamation cover but increase for the remaining soil covers (i.e., SWSS, SV10, SV27, and SV60). These results are not unexpected as, different from D3, the SWSS reclamation cover and three SV sites have low storage capabilities. Overall, the values of DS (typically 1–6 mm) are relatively small compared to other water balance components AET (on the order of hundreds of millimeters) and NP (on the order of from tens to hundreds of millimeters). Therefore, DS dynamics were not used here to compare the performance of reclamation covers with natural soil profiles.
i. Water balance components as a proportion of precipitation and PET
The proportions of AET and NP expressed as a percentage of precipitation (PPT) and PET are shown in Fig. 11 for the baseline and future periods. NP proportions for the sites with lower LAI values (e.g., SV10) are higher, and vice versa; these trends are reversed for the AET proportions. Little to no increase in future AET/PPT or AET/PET ratios compared to the baseline is evident, suggesting a linear response of AET to increases in either PPT or PET. In contrast, the increase in NP/PPT and NP/PET becomes larger particularly toward the end of the twenty-first century and for the higher-water-storage sites (e.g., reclamation covers).
The proportions of growing season (a) AET to precipitation, (b) NP to precipitation, (c) AET to PET, and (d) NP to PET during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods based on RCP2.6, RCP4.5, and RCP8.5 of CanESM2 for all five study sites.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The proportions of growing season (a) AET to precipitation, (b) NP to precipitation, (c) AET to PET, and (d) NP to PET during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods based on RCP2.6, RCP4.5, and RCP8.5 of CanESM2 for all five study sites.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The proportions of growing season (a) AET to precipitation, (b) NP to precipitation, (c) AET to PET, and (d) NP to PET during the baseline (1961–90) and future (2016–40, 2041–70, and 2071–2100) periods based on RCP2.6, RCP4.5, and RCP8.5 of CanESM2 for all five study sites.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The increase in future AET relative to the baseline is approximately the same as the increase in PET (results not shown) and, obviously, not the same as the increase in precipitation. The future precipitation increases faster than AET (shown in the probability distributions in Fig. 12 for precipitation and AET at all study sites). Consequently, the increase in future precipitation from the baseline results in increasing NP rates during the twenty-first century.
Rate of change in the probability distributions of growing season precipitation and AET during the baseline (dashed lines) and future periods (solid lines) based on three RCPs at each of the study sites. Green and black lines show the average probability distributions of all the future time periods and RCPs for precipitation and AET, respectively, reflecting the relative slopes at which future precipitation and AET are expected to change.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Rate of change in the probability distributions of growing season precipitation and AET during the baseline (dashed lines) and future periods (solid lines) based on three RCPs at each of the study sites. Green and black lines show the average probability distributions of all the future time periods and RCPs for precipitation and AET, respectively, reflecting the relative slopes at which future precipitation and AET are expected to change.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
Rate of change in the probability distributions of growing season precipitation and AET during the baseline (dashed lines) and future periods (solid lines) based on three RCPs at each of the study sites. Green and black lines show the average probability distributions of all the future time periods and RCPs for precipitation and AET, respectively, reflecting the relative slopes at which future precipitation and AET are expected to change.
Citation: Journal of Hydrometeorology 19, 11; 10.1175/JHM-D-17-0230.1
The NP rates for a cover system can be classified as “very low,” “low,” and “moderate” when the proportion of NP for a given year is between 1% and 5%, between 5% and 15%, and between 10% and 40% of precipitation for a given year, respectively (Ayres and O’Kane 2013). Based on this classification, the D3, SWSS, and SV60 sites have low rates, while SV10 and SV27 have moderate NP rates during the baseline period. With climate change, D3, SWSS, and SV60 might begin to shift from low to slightly moderate NP rates, while SV10 and SV27 remain in the moderate range (see Fig. 11b).
Although NP increases at all sites, the relative shifts in NP are greater at sites with more water storage. The higher rates of NP will lead to more water moving through the reclamation cover and into the underlying waste, ultimately passing through this waste to be released to downgradient surface water and groundwater. An increase in NP with time would result in more rapid flushing of mobile contaminants from the waste but at the cost of great chemical loading to the downgradient receptors. Increasing NP may also change the water content within the underlying waste, leading to changes in rates of weathering (e.g., oxidation).
4. Conclusions
This study utilized future climate change projections based on the most recent GCM outputs as input for physically based SVAT models of water balance for reclamation soil covers and natural soil profiles near Fort McMurray, Alberta, Canada. The downscaled temperature indicates increases for all RCPs during the twenty-first century, with a maximum increase of 138% and minimum increase of 28% (median values) compared to the baseline period. Overall, the growing season precipitation and PET are projected to increase irrespective of RCP and time period during the twenty-first century according to the Canadian GCM (CanESM2). Maximum increases in projected growing season precipitation and PET for RCP8.5 in 2071–2100 are 25.5% and 11.9%, respectively.
During the historical baseline period (1961–90), reclamation covers show higher AET and lower NP than natural sites. Overall, the important water balance components (i.e., AET and NP) of the reclamation covers and natural sites are expected to increase throughout the twenty-first century, regardless of the RCP, time period, or soil profile considered. However, the magnitude of the shifts in projected AET and NP are subject to the selection of RCP, time period, and study site. Greater increases in AET and NP are expected toward the end of the twenty-first century in response to increases in growing season PET and precipitation. When compared to shifts in future AET, changes in NP between baseline and future values for the sites with higher water storage (i.e., the two reclamation covers) are more dramatic. As a consequence, increases in NP for these sites will likely have a greater impact on adjacent water bodies because of proportionally greater water release and the concomitant release of constituents flushed from the mine waste (e.g., overburden or tailings).
The implications for higher future rates of NP are wide ranging. Increased rates of NP may lead to increases in base flow in surface water courses that feed wetlands or lakes. They also might lead to more rapid flushing of mobile chemical constituents present within the mine waste, with a resultant increase in chemical loading to these surface water receptors in the short term. Increases in NP could also result in elevated degrees of saturation, a rise in water pressure, and a concomitant rising water table within the mine waste. This may reduce weathering processes tied to oxygen availability but also result in an increased risk of geotechnical instability of the waste dumps and containment structures. Future studies should be undertaken in which this modeling approach is applied to an entire site-specific oil sands mine waste landform to evaluate some of these potential changes in landform performance.
The fine-textured reclamation covers placed over fine-textured mine waste (e.g., South Bison Hill) are designed to maximize the storage of plant-available water and consequently have relatively low rates of NP. However, the fine-textured mine waste also contains large concentrations of readily soluble salts (e.g., Appels et al. 2017) that will be released to adjacent surface waters because of flushing by water ingress into, and subsequently out of, the dump. The SWSS reclamation cover is designed in a similar manner with fine-textured cover soils placed over relatively free-draining sand tailings to maximize plant-available water for revegetation efforts. However, in this case, large increases in NP might have a positive influence on mine closure by flushing mobile salts from the sand tailings more rapidly, thereby reducing long-term loading from the reclaimed site to surface waters (Booterbaugh et al. 2015).
Multiple (100) realizations of LARS-WG were integrated into HYDRUS-1D to encompass the uncertainty in the water balance components due to natural climate variability, with the estimation of uncertainty due to the use of three RCPs and three future time periods. Overall, the uncertainty ranges (95th minus 5th percentile value) of both AET and NP increase in the future relative to the baseline but vary depending on the RCP and time period. The increases in future growing season median AET and NP relative to baseline values are also subject to uncertainty because of the selection of RCP and time period. Further research is in progress to include multiple GCMs with the RCPs to incorporate uncertainties in the simulations of water balance components at the reclamation covers and natural sites with spatial and temporal variability in vegetation, soil properties, and topographical settings.
Acknowledgments
We acknowledge funding provided by Syncrude Canada, Ltd., the Natural Sciences and Engineering Research Council (NSERC) of Canada (File 428588-11), and the Department of Civil, Geological and Environmental Engineering, University of Saskatchewan, Canada. The first author also acknowledges financial support from the Government of Saskatchewan in the form of an SK Innovation and Opportunity Scholarship. The first author thanks Dr. Willemijn Appels for her assistance in creating a batch controller for HYDRUS-1D using MATLAB. We thank Stephanie Villeneuve for designing Fig. 1.
APPENDIX A
Downscaling Methods
a. LARS-WG
The statistics S and the monthly mean of maximum and minimum temperature for RCPs during the future periods are perturbed using the observed monthly statistics 
Relative change factors for CanESM2 based on RCP2.6 during 2071–2100 used in LARS-WG.
b. Delta change method
APPENDIX B
VG Equations
a. Single-porosity model
b. Dual-porosity model
c. Parameters of HYDRUS-1D
The VG parameters for the mobile region (i.e., θrm, θsm, αm, nm) and for the immobile region (i.e., θrim, θsim, αim, nim) in Table B1, or for the single region (i.e., θr, θs, α, n) in Table B2, are used to characterize the soil water retention curve (WRC) for the respective region, while Ks is used to define the hydraulic conductivity function K(h) of the soil profile in HYDRUS-1D. Tables B1 and B2 show the calibrated and validated parameters used to simulate long-term water balance of the respective oil sands mine reclamation covers.
VG parameters and saturated hydraulic conductivity for the two SWSS cover materials and three D3 cover materials.
Statistics of the estimated and optimized VG parameters and saturated hydraulic conductivity for the three SV sites (adapted from Huang et al. 2011b).
REFERENCES
Alam, M. S., and A. Elshorbagy, 2015: Quantification of the climate change-induced variations in intensity-duration-frequency curves in the Canadian prairies. J. Hydrol., 527, 990–1005, https://doi.org/10.1016/j.jhydrol.2015.05.059.
Allen, R. G., L. S. Pereira, D. Raes, and M. Smith, 1998: Crop evapotranspiration—Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, 300 pp.
AMEC Earth and Environmental and Paragon Soil and Environmental Consulting, Inc., 2005: Results from long term vegetation plots established in the oil sands region. Cumulative Environmental Management Association Oil Sands Soil and Vegetation Working Group Rep., 65 pp.
Appels, W. M., S. N. Wall, S. L. Barbour, M. J. Hendry, C. F. Nichol, and S. R. Chowdhury, 2017: Pyrite weathering in reclaimed shale overburden at an oil sands mine near Fort McMurray, Canada. Mine Water Environ., 36, 479–494, https://doi.org/10.1007/s10230-017-0454-4.
Ayres, B., and M. O’Kane, 2013: Design, construction, and performance of closure cover systems for spent heap leach piles—A state-of-the-art review. Proc. Int. Heap Leach Conf., Vancouver, BC, Canada, The University of British Columbia, 240–251.
Barbour, S. L., and Coauthors, 2004: Tracking the evolution of reclaimed landscapes through the use of instrumented watersheds—A brief history of the Syncrude Southwest 30 Overburden Reclamation Research Program. Proc. Int. Instrumented Watershed Symp., Edmonton, AB, Canada, Syncrude Canada Limited and Natural Sciences and Engineering Research Council of Canada.
Barr, A. G., G. van der Kamp, T. A. Black, J. H. McCaughey, and Z. Nesic, 2012: Energy balance closure at the BERMS flux towers in relation to the water balance of the White Gull Creek watershed 1999–2009. Agric. Meteor., 153, 3–13, https://doi.org/10.1016/j.agrformet.2011.05.017.
Beckingham, J. D., and J. H. Archibald, 1996: Field guide to ecosites of northern Alberta. Natural Resources Canada Special Rep. 5, 516 pp.
Boese, C. D., 2003: The design and installation of a field instrumentation program for the evaluation of soil–atmosphere water fluxes in a vegetated cover over saline/sodic shale overburden. M.S. thesis, Dept. of Civil Engineering, University of Saskatchewan, 170 pp., http://hdl.handle.net/10388/etd-12172012-083151.
Booterbaugh, A. P., R. B. Laurence, and C. A. Mendoza, 2015: Geophysical characterization of an undrained dyke containing an oil sands tailings pond, Alberta, Canada. J. Environ. Eng. Geophys., 20, 303–317, https://doi.org/10.2113/JEEG20.4.303.
Braden, H., 1985: Ein Energiehaushalts- und Verdunstungsmodell for Wasser und Stoffhaushaltsuntersuchungen landwirtschaftlich genutzer Einzugsgebiete. Mittelungen Dtsch. Bodenkd. Geselschaft, 42, 294–299.
Brutsaert, W., 1982: Evaporation into Atmosphere: Theory, History, and Applications. D. Reidel, 302 pp.
CAPP, 2016: Canada’s oil sands. CAPP Rep., 31 pp., http://www.oscaalberta.ca/wp-content/uploads/2017/06/CAPP-Oil-Sands-Fact-Booklet.pdf.
Carrera-Hernández, J. J., C. A. Mendoza, K. J. Devito, R. M. Petrone, and B. D. Smerdon, 2011: Effects of aspen harvesting on groundwater recharge and water table dynamics in a subhumid climate. Water Resour. Res., 47, 1–18, https://doi.org/10.1029/2010WR009684.
CEMA, 2006: Field Manual for Land Capability Determination. Vol. 1, Land Capability Classification System for Forest Ecosystems in the Oil Sands, 3rd ed., Alberta Environment, 148 pp., http://www.assembly.ab.ca/lao/library/egovdocs/2006/alen/158348.pdf.
Chasmer, L., H. McCaughey, A. Barr, A. Black, A. Shashkov, P. Treitz, and T. Zha, 2008: Investigating light-use efficiency across a jack pine chronosequence during dry and wet years. Tree Physiol., 28, 1395–1406, https://doi.org/10.1093/treephys/28.9.1395.
Dobchuk, B. S., R. E. Shurniak, S. L. Barbour, M. A. O’Kane, and Q. Song, 2013: Long-term monitoring and modelling of a reclaimed watershed cover on oil sands tailings. Int. J. Min. Reclam. Environ., 27, 180–201, https://doi.org/10.1080/17480930.2012.679477.
Elshorbagy, A., and S. L. Barbour, 2007: Probabilistic approach for design and hydrologic performance assessment of reconstructed watersheds. J. Geotech. Geoenviron. Eng., 133, 1110–1118, https://doi.org/10.1061/(ASCE)1090-0241(2007)133:9(1110).
Elshorbagy, A., A. Jutla, and J. Kells, 2007: Simulation of the hydrological processes on reconstructed watersheds using system dynamics. Hydrol. Sci. J., 52, 538–562, https://doi.org/10.1623/hysj.52.3.538.
Elshorbagy, A., A. Nazemi, and M. S. Alam, 2015: Analyzing the variations in intensity-duration-frequency (IDF) curves in the city of Saskatoon under climate change. University of Saskatchewan Centre for Advanced Numerical Simulation Series Rep. CAN-15-01, 167 pp.
Feddes, R. A., E. Bresler, and S. P. Neuman, 1974: Field test of a modified numerical model for water uptake by root systems. Water Resour. Res., 10, 1199–1206, https://doi.org/10.1029/WR010i006p01199.
Franczyk, J., and H. Chang, 2009: The effects of climate change and urbanization on the runoff of the Rock Creek basin in the Portland metropolitan area, Oregon, USA. Hydrol. Processes, 23, 805–815, https://doi.org/10.1002/hyp.7176.
Gerke, H. H., and M. T. van Genuchten, 1993: Evaluation of a first-order water transfer term for variably saturated dual-porosity flow models. Water Resour. Res., 29, 1225–1238, https://doi.org/10.1029/92WR02467.
Githui, F., W. Gitau, F. Mutua, and W. Bauwens, 2009: Climate change impact on SWAT simulated streamflow in western Kenya. Int. J. Climatol., 29, 1823–1834, https://doi.org/10.1002/joc.1828.
Hall, F. G., Y. E. Shimabukuro, and K. F. Huemmrich, 1995: Remote sensing of forest biophysical structure using mixture decomposition and geometric reflectance models. Ecol. Appl., 5, 993–1013, https://doi.org/10.2307/2269350.
Hashmi, M. Z., A. Y. Shamseldin, and B. W. Melville, 2011: Comparison of LARS-WG and SDSM for simulation and downscaling of extreme precipitation events in a watershed. Stochastic Environ. Res. Risk Assess., 25, 475–484, https://doi.org/10.1007/s00477-010-0416-x.
Hassanzadeh, E., A. Nazemi, and A. Elshorbagy, 2014: Quantile-based downscaling of precipitation using genetic programming: Application to IDF curves in Saskatoon. J. Hydrol. Eng., 19, 943–955, https://doi.org/10.1061/(ASCE)HE.1943-5584.0000854.
Howard, E. A., S. T. Gower, J. A. Foley, and C. J. Kucharik, 2004: Effects of logging on carbon dynamics of a jack pine forest in Saskatchewan, Canada. Global Change Biol., 10, 1267–1284, https://doi.org/10.1111/j.1529-8817.2003.00804.x.
Huang, M., A. Elshorbagy, S. L. Barbour, J. Zettl, and B. C. Si, 2011a: System dynamics modeling of infiltration and drainage in layered coarse soil. Can. J. Soil Sci., 91, 185–197, https://doi.org/10.4141/cjss10009.
Huang, M., S. L. Barbour, A. Elshorbagy, J. Zettl, and B. C. Si, 2011b: Water availability and forest growth in coarse-textured soils. Can. J. Soil Sci., 91, 199–210, https://doi.org/10.4141/cjss10012.
Huang, M., S. L. Barbour, A. Elshorbagy, J. Zettl, and B. C. Si, 2011c: Infiltration and drainage processes in multi-layered coarse soils. Can. J. Soil Sci., 91, 169–183, https://doi.org/10.4141/cjss09118.
Huang, M., S. L. Barbour, and S. K. Carey, 2015: The impact of reclamation cover depth on the performance of reclaimed shale overburden at an oil sands mine in northern Alberta, Canada. Hydrol. Processes, 29, 2840–2854, https://doi.org/10.1002/hyp.10229.
Huntington, T. G., 2006: Evidence for intensification of the global water cycle: Review and synthesis. J. Hydrol., 319, 83–95, https://doi.org/10.1016/j.jhydrol.2005.07.003.
IPCC, 2000: Land Use, Land-Use Change and Forestry: A Special Report of the IPCC. R. Watson et al., Eds., Cambridge University Press, 30 pp.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
Keshta, N., A. Elshorbagy, and S. Carey, 2009: A generic system dynamics model for simulating and evaluating the hydrological performance of reconstructed watersheds. Hydrol. Earth Syst. Sci., 13, 865–881, https://doi.org/10.5194/hess-13-865-2009.
Keshta, N., A. Elshorbagy, and S. Carey, 2012: Impacts of climate change on soil moisture and evapotranspiration in reconstructed watersheds in northern Alberta, Canada. Hydrol. Processes, 26, 1321–1331, https://doi.org/10.1002/hyp.8215.
Kienzle, S. W., M. W. Nemeth, J. M. Byrne, and R. J. MacDonald, 2012: Simulating the hydrological impacts of climate change in the upper North Saskatchewan River basin, Alberta, Canada. J. Hydrol., 412–413, 76–89, https://doi.org/10.1016/j.jhydrol.2011.01.058.
Kuusisto, E., 1980: On the values and variability of degree-day melting factor in Finland. Nord. Hydrol., 11, 235–242, https://doi.org/10.2166/nh.1980.0011.
Lavigne, M. B., R. J. Foster, G. Goodine, P. Y. Bernier, and C. H. Ung, 2005: Alternative method for estimating aboveground net primary productivity applied to balsam fir stands in eastern Canada. Can. J. For. Res., 35, 1193–1201, https://doi.org/10.1139/x05-052.
Leta, O. T., A. I. El-Kadi, H. Dulai, and K. A. Ghazal, 2016: Assessment of climate change impacts on water balance components of Heeia watershed in Hawaii. J. Hydrol. Reg. Stud., 8, 182–197, https://doi.org/10.1016/j.ejrh.2016.09.006.
Mango, L. M., A. M. Melesse, M. E. McClain, D. Gann, and S. G. Setegn, 2011: Land use and climate change impacts on the hydrology of the upper Mara River basin, Kenya: Results of a modeling study to support better resource management. Hydrol. Earth Syst. Sci., 15, 2245–2258, https://doi.org/10.5194/hess-15-2245-2011.
Najafi, M. R., H. Moradkhani, and I. W. Jung, 2011: Assessing the uncertainties of hydrologic model selection in climate change impact studies. Hydrol. Processes, 25, 2814–2826, https://doi.org/10.1002/hyp.8043.
OKC, 2001: Southwest sand storage and 30-dump automated water balance monitoring systems at Syncrude Canada Ltd. OKC Rep. 653-2, 19 pp.
OKC, 2007: Soil-atmosphere field response modelling of Southwest Sand Storage facility cell 32. OKC Rep. 690/14-01, 16 pp.
OKC, 2016: Instrumented watershed monitoring program at the Southwest Sand Storage facility: Performance monitoring report for the period January 2015 to December 2015. OKC Rep. 690/01-72, 26 pp.
Pan, S., and Coauthors, 2015: Responses of global terrestrial evapotranspiration to climate change and increasing atmospheric CO2 in the 21st century. Earth’s Future, 3, 15–35, https://doi.org/10.1002/2014EF000263.
Price, J. S., R. G. McLaren, and D. L. Rudolph, 2010: Landscape restoration after oil sands mining: Conceptual design and hydrological modelling for fen reconstruction. Int. J. Min. Reclam. Environ., 24, 109–123, https://doi.org/10.1080/17480930902955724.
Prudhomme, C., N. Reynard, and S. Crooks, 2002: Downscaling of global climate models for flood frequency analysis: Where are we now? Hydrol. Processes, 16, 1137–1150, https://doi.org/10.1002/hyp.1054.
Qualizza, C., D. Chapman, S. L. Barbour, and B. Purdy, 2004: Reclamation research at Syncrude Canada’s mining operation in Alberta’s Athabasca oil sands region. Proc. 16th Int. Conf. on Ecological Restoration, Victoria, BC, Canada, Society for Ecological Restoration.
Racsko, P., L. Szeidl, and M. Semenov, 1991: A serial approach to local stochastic weather models. Ecol. Modell., 57, 27–41, https://doi.org/10.1016/0304-3800(91)90053-4.
Rogelj, J., M. Meinshausen, and R. Knutti, 2012: Global warming under old and new scenarios using IPCC climate sensitivity range estimates. Nat. Climate Change, 2, 248–253, https://doi.org/10.1038/nclimate1385.
Rooney, R. C., D. T. Robinson, and R. Petrone, 2015: Megaproject reclamation and climate change. Nat. Climate Change, 5, 963–966, https://doi.org/10.1038/nclimate2719.
Rosenzweig, M. L., 1968: Net primary productivity of terrestrial communities: Prediction from climatological data. Amer. Nat., 102, 67–74, https://doi.org/10.1086/282523.
Schindler, D. W., and W. F. Donahue, 2006: An impending water crisis in Canada’s western prairie provinces. Proc. Natl. Acad. Sci. USA, 103, 7210–7216, https://doi.org/10.1073/pnas.0601568103.
Schneider, R. R., 2013: Alberta’s natural subregions under a changing climate: Past, present, and future. Alberta Biodiversity Monitoring Institute Rep., 80 pp.
Semenov, M. A., and E. M. Barrow, 1997: Use of a stochastic weather generator in the development of climate change scenarios. Climatic Change, 35, 397–414, https://doi.org/10.1023/A:1005342632279.
Semenov, M. A., and E. M. Barrow, 2002: LARS-WG: A stochastic weather generator for use in climate impact studies. Rothamsted Research Rep., 28 pp., http://resources.rothamsted.ac.uk/sites/default/files/groups/mas-models/download/LARS-WG-Manual.pdf.
Simunek, J., N. J. Jarvis, M. T. van Genuchten, and A. Gardenas, 2003: Review and comparison of models for describing nonequilibrium and preferential flow and transport in the vadose zone. J. Hydrol., 272, 14–35, https://doi.org/10.1016/S0022-1694(02)00252-4.
Simunek, J., M. T. van Genuchten, and M. Sejna, 2012: The HYDRUS software package for simulating the two- and three-dimensional movement of water, heat, and multiple solutes in variably-saturated porous media. University of California, Riverside, Rep., 230 pp.
Simunek, J., M. Sejna, H. Saito, M. Sakai, and M. T. van Genuchten, 2013: The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat, and multiple solutes in variably-saturated media. University of California, Riverside, Rep., 305 pp.
Srivastav, R. K., A. Schardong, and S. P. Simonovic, 2014: Equidistance quantile matching method for updating IDF curves under climate change. Water Resour. Manage., 28, 2539–2562, https://doi.org/10.1007/s11269-014-0626-y.
Straker, J., M. O’Kane, S. Carey, T. E. Baker, D. Charest, and R. Shurniak, 2014: Towards an ecohydrologic classification of reclaimed watersheds: Methods for estimating soil water regime on reclaimed mine waste materials; and relationships between reclamation and surface water balances in Teck’s reclaimed coal-mining watersheds. 38th Annual British Columbia Mine Reclamation Symp., Prince George, BC, Canada, British Columbia Technical and Research Committee on Reclamation, https://doi.org/10.14288/1.0042684.
Strong, W. L., and K. R. Leggat, 1981: Ecoregions of Alberta. Alberta Energy and Natural Resources Tech. Rep. T/4, 64 pp.
Suncor Energy, Inc., 2007: Climate change in the oil sands region. Voyager South Project Environmental Impact Rep., 134 pp.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.
Thompson, C., C. A. Mendoza, and K. J. Devito, 2017: Potential influence of climate change on ecosystems within the Boreal Plains of Alberta. Hydrol. Processes, 31, 2110–2124, https://doi.org/10.1002/hyp.11183.
van Genuchten, M. T., 1980: A closed-form equation for predicting the hydraulic conductivity of unmatured soils. Soil. Sci. Soc. Amer. J., 44, 892–898, https://doi.org/10.2136/sssaj1980.03615995004400050002x.
Vogel, J. G., and S. T. Gower, 1998: Carbon and nitrogen dynamics of boreal jack pine stands with and without a green alder understory. Ecosystems, 1, 386–400, https://doi.org/10.1007/s100219900032.
Zettl, J., S. L. Barbour, M. Huang, B. Si, and L. A. Leskiw, 2011: Influence of textural layering on field capacity of coarse soils. Can. J. Soil Sci., 91, 133–147, https://doi.org/10.4141/cjss09117.